How policy influence varies with race and gender in the US courts of appeals

Abstract

Even when women and people of color achieve positions of political power, institutional norms may combine with social constructions of difference to create a system in which power is distributed disproportionately. Such a pattern is evident in the US courts of appeals. Each case is resolved by a panel of three judges who also decide whether the opinion should be binding precedent (i.e., published) or not. I theorized that the variety of views and extended deliberation often attributed to diversity in a small-group environment depressed the rate of publication if judges were willing to compromise on the outcome but less willing to publish an opinion after such compromise. Using a massive original dataset of virtually all dispositive circuit opinions from 2002 to 2012, I found that homogeneous panels (98% of which are composed of white men) shaped policy more frequently than diverse panels.

Introduction

Women and people of color achieving positions of political power does not guarantee that they will be able to wield the same amount of power as similarly situated white men. Institutional norms may combine with social constructions of difference to distribute power disproportionately (see Karpowitz et al., 2012; Mendelberg et al., 2014). Such a pattern is evident in the US courts of appeals. Benign procedural practices and laudable deliberative processes combine with divergent viewpoints generated by fundamentally different social experiences to create a system in which power is exercised unequally.

Judges on federal circuit courts split their time between resolving important, policy-shaping cases and routine matters that do not have an impact on circuit law (Tillman and Hinkle, 2018). In order to keep up with an increasing volume of appeals, circuit courts developed the practice of designating some opinions as unpublished (Hazelton et al., 2016). In future cases, judges are under no obligation to follow these rulings, which means they do not play a role in shaping policy. Each case heard in the circuit courts is resolved by a panel of three judges who decide whether the opinion should be published, and, therefore, binding precedent. This procedural step provides an opportunity for compromise. Judge Wald of the District of Columbia Circuit wrote the following:

I have seen judges purposely compromise on an unpublished decision incorporating an agreed-upon result in order to avoid a time-consuming public debate about what law controls. I have even seen wily would-be dissenters go along with a result they do not like so long as it is not elevated to a precedent (Wald, 1995: 1374).

Research shows that diversity in decision-making groups produces a wider range of ideas and deliberation (Haire et al., 2013; McLeod et al., 1996). This dynamic can lead to higher quality outputs. For example, Sommers (2006) demonstrates that racially diverse juries make fewer legal errors than homogeneous juries. Relatedly, Clayton et al. (2019) show that policies formulated by diverse decision-making groups can have greater legitimacy than those made by homogeneous groups. However, the very fact that diversity generates a wider range of ideas may inadvertently depress the ability of diverse groups of judges to shape legal policy, resulting in law disproportionately crafted by homogeneous groups of judges. Since more diverse groups tend to be more likely to produce a wider range of perspectives, they are also more likely to generate a need to compromise. Many forms of compromise are available to circuit judges who do not initially see eye-to-eye about the resolution of a case. A key way to compromise is to leave the opinion unpublished so that the judge unhappy with the result can be assured it will not shape the law of the circuit (Hazelton et al., 2016). Consequently, cases where there is more likely to be disagreement among judges may also be more likely to result in unpublished opinions. This, in turn, suggests that judicial diversity may be associated with lower levels of opinion publication and, thus, policy influence.

I tested this theory using a comprehensive original dataset of circuit opinions from 2002 to 2012 to explore the correlation between publication and panel diversity (n = 233,931). There was consistent evidence of demographic diversity depressing publication rates. Overall, cases with homogeneous panels published 28.1% of their opinions while panels with race and gender diversity published 19.3%. Regression analysis accounting for a number of other factors (including ideological diversity) confirmed the important role of demographic diversity.

The publication gap looms large when considering that the circuit courts resolve more than twenty thousand cases per year. The estimated diversity gap represents hundreds of opinions annually. This indicates that federal law is shaped by homogeneous groups of judges to a greater degree than diverse groups of judges. As a result, groups with a larger number of representatives on the federal courts have disproportionate power since they are statistically more likely to be part of a homogeneous decision-making group than their less well-represented colleagues.

Publication and policy

Since the US Supreme Court hears such a small number of cases each year, the vast majority of federal case law is shaped by judges at the circuit level (Haire and Moyer, 2015). But circuit judges do not spend all of their time wielding the power to shape policy. They must also dispense with the important tasks of error correction and ensuring perceptions of procedural fairness. While resolving all appeals brought before them – including repetitive, routine, and even frivolous legal arguments – is essential for maintaining justice, the result is that only a fraction of the cases a circuit judge works on offers the chance to shape policy (Hazelton et al., 2016). This institutional framework creates the possibility that systemic racism and sexism within society could operate in subtle and difficult to detect ways by reducing the number of cases in which female judges or judges of color exercise power over policy.

Publication is a key element of power. Legal doctrine, often formalized in court rules, ¹ mandates that only opinions that a panel designates as “published” are binding precedent, that is, they are required to be applied throughout the circuit (Hazelton et al., 2016). Furthermore, research has established that binding precedents have far more impact in the circuit courts than those judges are not required to apply (Hinkle, 2015). Consequently, examining what factors play a role in the publication decision is vital for understanding the distribution of power among circuit judges. Key to this decision is the fact that it is made in a group context by a panel of three judges (Bowie et al., 2014). There is substantial variation in the composition of these panels because panel assignments rotate on a regular basis (Hazelton et al., 2016). Panels can be composed of three judges of the same gender and race or three judges with varying demographic backgrounds. There is also a wide range of ideological diversity. Research indicates that the presence of diversity on a panel can shape case outcomes in at least some issue areas (Boyd et al., 2010; Farhang and Wawro, 2004; Harris and Sen, 2019; Kastellec, 2013) and increase the number of legal issues discussed in an opinion (Haire et al., 2013). In a small-group decision-making environment the makeup of the group matters. Homogeneity can lead to different outputs than heterogeneity. I posited that panel diversity plays a role in publication as well. ²

Diversity, difference, and deliberation

Within a group setting, differences in background, perspective, and experiences frequently generate a wider range of ideas and, thus, greater deliberation, than a homogeneous group (see McLeod et al., 1996; Moyer et al., 2020; Sommers, 2006). McLeod et al. (1996) found that racially diverse groups given a brainstorming task produced ideas that were more effective and feasible than those produced by homogeneous groups. In the courts context, there is evidence that circuit panels with a wider range of ideological beliefs deliberate more since they take a longer time to resolve cases than ideologically homogeneous panels (Bowie et al., 2014: 110). Demographic diversity is likely to operate in a similar fashion. Men and women can have different perspectives based on their role in society, and, therefore, may bring different information and priorities to policy deliberations (Boyd et al., 2010; Fix and Johnson, 2017; Tillman and Hinkle, 2018). Similarly, and perhaps to an even greater extent, a person’s race also shapes how they have experienced society and their view of the role of law within that society (Sommers, 2006; Tillman and Hinkle, 2018). Some of these different views manifest in terms of women and people of color tending to be more liberal. But breadth and variation in perspectives can extend well beyond differences in ideology.

This variation in perspectives is one reason why diversity is sought as a desirable feature of group dynamics. In the context of any democratic institution, having a variety of perspectives that reflects those of citizens has an important legitimizing function (Clayton et al., 2019; Fix and Johnson, 2017; Hayes and Hibbing, 2017). Moreover, just as multiple judges are used to increase the probability of reaching the correct outcome (Haire et al., 2013), so multiple perspectives may be valuable to achieve the same goal. In spite of these normative goals, the very diversity that can strengthen the deliberative process may also have the counterintuitive result of disproportionately distributing policymaking power. As Mendelberg et al. (2014: 35) point out, “Representation depends not only on gender composition, but also on institutional norms and procedures that are neutral on their face but carry profound consequences for social inequality.”

A small group tasked with reaching a decision in the face of diverse viewpoints often relies upon compromise. ³ This is cited as a strength diverse groups provide over homogeneous groups (Sommers, 2006). For panels there are two dimensions along which compromise is possible. First, they may compromise over the substance of the opinion. Second, the publication decision offers a point of compromise (Hazelton et al., 2016; Wald, 1995). ⁴

A panel often makes the publication decision in combination with the decision of how to resolve the case (Bowie et al., 2014). ⁵ This concurrent timing facilitates negotiation. In broadly stylized terms for purposes of exposition, each judge’s ideal result would be their preferred outcome with a published opinion that constitutes the law of the circuit. However, in the face of divergent views a judge may be willing to agree to the opinion being unpublished as long as they obtain their desired outcome. Or a judge may agree to a compromise regarding the outcome. In the case of a compromise outcome, both sides are likely to prefer an unpublished opinion since neither is fully comfortable with the resolution and, therefore, may not want to make it binding precedent. Finally, the least optimal compromise solution (without losing completely) is to agree to the other side’s preferred outcome, but insist on the opinion being unpublished. That at least limits the damage to the current case. In summary, a judge’s individual utility from the various possible combinations of panel decisions can be characterized as follows:

Preferred outcome, Published

>

Preferred outcome, Unpublished

>

Compromise outcome, Unpublished

>

Opposed outcome, Unpublished

>

Opposed outcome, Published

Many factors influence which precise types of compromise are more or less likely in any given case. That nuanced dynamic is beyond the scope of this study. Rather, I focused on the implications diversity has for the likelihood of publishing an opinion. If all panel members agree on the case outcome (and the case is a type generally considered for publication), then the panel is incentivized to publish the opinion and shape the future law of the circuit. Conversely, if a panel is forging a compromise, the most logical compromise options involve leaving a ruling unpublished. While a panel could publish a compromise outcome, that would be odd since neither side should want to establish a sub-optimal result as law. We know from empirical research that even in published cases a two-judge majority sometimes allows the one minority judge to secure their preferred outcome (e.g., Boyd et al., 2010; Kastellec, 2013). Logic would suggest that more balanced forms of compromise are even more common. As a result, panels that must work to a compromise should be less likely to publish the opinion.

It is impossible to know which panels generate consensus in a case from the beginning and which require extensive deliberation and compromise. But it is plausible that diverse groups are more likely to generate disagreement than homogeneous groups. Therefore, I hypothesized that panels that are diverse in terms of gender, race or ethnicity, or ideology will be less likely to publish their opinion compared with panels that are homogeneous on those dimensions.

Data and results

In order to evaluate whether panel diversity is related to publication decisions, I utilized an original dataset of virtually all opinions issued by US courts of appeals from 2002 to 2012. ⁶ I focused exclusively on three-judge panel opinions. The unit of analysis was the opinion (n = 233,931). The outcome variable equals 1 if a ruling is published (and zero otherwise). In the full dataset, 23% of opinions were published. Since the outcome variable is dichotomous, I estimated a probit model.

The primary explanatory variables measured the level of panel diversity. I looked at whether a panel had gender diversity, racial diversity, both, or neither by using three indicator variables. The first equals 1 when a panel has at least one man and one woman but all panel members share the same race or ethnicity. The second indicator equals 1 when panel members have different racial or ethnic backgrounds but are all the same gender. Finally, the third variable equals 1 for panels that are diverse in terms of both race and gender. The excluded category was panels that were homogeneous in terms of both gender and race. ⁷ Just over a quarter of the cases were resolved by such homogeneous panels, and 98% of homogeneous panels were composed of three white men. Figure 1 illustrates that all four possible types of panels using this typology of demographic diversity are well represented in the data. Figure 1 additionally shows the distribution of the outcome variable within each category.

Figure 1.

Distribution of opinions in terms of the panel’s demographic diversity and publication status.

It is important to account for ideological diversity as well, especially because demographic differences can be intertwined with ideological differences. Figure 2 illustrates the distribution of ideology for individual judges by race and gender using Judicial Common Space (JCS) scores. ⁸ While far from determinative, judges from different demographic groups do have somewhat different ideological dispositions. In order to measure the ideological diversity on a panel, I computed the average distance between the JCS scores of each pairwise combination of judges on the panel.

Figure 2.

Distribution of individual judge’s ideology.

Finally, I controlled for a range of case and circuit characteristics that were likely to influence publication and may be correlated with panel diversity. Amicus participation, oral argument, publication of the opinion being appealed, and a lack of unanimity all suggest the type of case where publication may be merited. Novelty as well as importance indicates publication is likely. For novel legal issues there are, by definition, no binding precedents available to cite. Consequently, I controlled for legal novelty by using the proportion of citations in an opinion that were to non-binding precedents. When the lower court is affirmed, an opinion is left unsigned (i.e., a per curiam opinion), or a simple order is issued in lieu of a full written opinion the court is less likely to publish. Publication may also vary by issue area ⁹ or circuit caseload, rules, or general tendency to publish. Appendix B.2 provides a detailed description of all these variables.

Figure 3 presents the results of the model. ¹⁰ As hypothesized, diverse panels were less likely to publish their opinions. ¹¹ This was true of both demographic and ideological diversity. Since a large dataset like this makes it possible to find statistically significant effects that are substantively quite small, I focused on the size of these effects. Figure 4 shows the predicted probability of publication for each type of demographic panel under the conditions of both ideological homogeneity and diversity. ¹² Ideologically homogeneous panels with both race and gender diversity published 3.7% fewer of their opinions compared with unified homogeneous panels. For ideologically diverse panels that disparity was a very similar 3.9%. Across all types of demographic panels, ideologically diverse panels were at least 4.5% less likely to publish compared with an ideologically homogeneous panel of the same demographic type.

Figure 3.

Probit regression estimates (and their 95% confidence intervals) of the effect of panel diversity and a range of control variables on the decision to publish.

Figure 4.

Predicted probability of publication for each demographic type of panel when (a) ideologically homogeneous and (b) diverse while holding all controls at their median value among published opinions. Ideological diversity is set to its 5% value for homogeneous panels and 95% value for diverse panels.

Implications

The benefits of diversity are frequently, and justifiably, extolled. Yet those benefits can have unintended consequences. Diversity can lead to stronger decisions because a wider variety of information and perspectives are brought to the table. Such variety increases the need for compromise. Indeed, forging results through compromise may also be seen as a normatively preferable outcome. But, taken together with the institutional feature that gives judges control over whether each opinion makes policy, the very benefit of driving compromise results could also decrease the extent to which diverse panels take advantage of their power to shape legal policy. Non-publication is a key tool of compromise. The empirical results bore out the logical implications. Diverse panels published fewer opinions.

What do these results mean for judicial institutions? First, they indicate that representation in the judiciary at levels similar to the population will not fully empower judges to influence policy on a par with their counterparts. Doing away with the publication norm is likely to cause more harm than it prevents since circuit courts face massive caseloads. The substantial increase in judicial staffing required to shift away from the practice of leaving some opinions unpublished is unlikely to be practicable or politically feasible. One possible solution could be to mandate a minimum level of diversity on each panel, although that would pose challenges and feasibility concerns as well. Although the solution is far from clear, the problem is. Power was exercised disproportionately among judges depending on their race and gender. This discrepancy emerged even after controlling for the effects of ideological differences among judges.

This study reinforced Karpowitz et al.’s (2012) point that institutions play a role in how demographic representation of historically marginalized groups is translated into influence. Gender norms and race relations can still reduce the authority and power of women and minorities relative to their similarly situated white, male peers. One of the ways this can manifest is through distribution of different types of work. The distinction between binding policymaking and completing tasks that have a lesser impact is not the sole purview of the judicial system. For example, some university committees serve as merely advisory bodies while others must approve any policy change before it happens. Many decisions university committees make on everything from promotion and tenure to academic dishonesty are merely advisory to a lone administrator. If the administrators and members of committees with actual power to directly shape policy decisions are more likely to be male and white than those serving in an advisory capacity, the same issue revealed in this study may exist. A seat at the table matters. But making sure everyone at the table has an equal voice matters too. We must look not only at the representation of women and minorities in policymaking positions, but also at how institutional rules and task distribution shape the power they exert.

While this study investigates a key area of inquiry, much work remains to be done. Diversity exists along several dimensions and can be operationalized in many ways. The interactions between systemic social dynamics and seemingly benign institutional structures are likely to vary in interesting ways in different countries, time periods, and institutions. Subtle inequalities cannot be addressed until they are understood. The results here point the way to a rich line of inquiry that deserves continued attention.

Supplemental Material

sj-pdf-1-rap-10.1177_20531680211029432 – Supplemental material for How policy influence varies with race and gender in the US courts of appeals

Supplemental material, sj-pdf-1-rap-10.1177_20531680211029432 for How policy influence varies with race and gender in the US courts of appeals by Rachael K. Hinkle in Research & Politics